Dynamic Convolution: Attention Over Convolution Kernels [PDF]
2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020 CVPR 2020 (Oral)
Chen, Yinpeng +5 more
openaire +4 more sources
DEA-Net: Single Image Dehazing Based on Detail-Enhanced Convolution and Content-Guided Attention [PDF]
IEEE Transactions on Image Processing, 2023 Single image dehazing is a challenging ill-posed problem which estimates latent haze-free images from observed hazy images. Some existing deep learning based methods are devoted to improving the model performance via increasing the depth or width of ...
Zixuan Chen, Zewei He, Zhe-ming Lu
semanticscholar +1 more source
Channel-wise Topology Refinement Graph Convolution for Skeleton-Based Action Recognition [PDF]
IEEE International Conference on Computer Vision, 2021 Graph convolutional networks (GCNs) have been widely used and achieved remarkable results in skeleton-based action recognition. In GCNs, graph topology dominates feature aggregation and therefore is the key to extracting representative features.
Yuxin Chen +5 more
semanticscholar +1 more source
UniFormer: Unifying Convolution and Self-Attention for Visual Recognition [PDF]
IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022 It is a challenging task to learn discriminative representation from images and videos, due to large local redundancy and complex global dependency in these visual data.
Kunchang Li +7 more
semanticscholar +1 more source
Conformer: Convolution-augmented Transformer for Speech Recognition [PDF]
Interspeech, 2020 Recently Transformer and Convolution neural network (CNN) based models have shown promising results in Automatic Speech Recognition (ASR), outperforming Recurrent neural networks (RNNs).
Anmol Gulati +10 more
semanticscholar +1 more source
Omni-Dimensional Dynamic Convolution [PDF]
International Conference on Learning Representations, 2022 Learning a single static convolutional kernel in each convolutional layer is the common training paradigm of modern Convolutional Neural Networks (CNNs).
Chao Li, Aojun Zhou, Anbang Yao
semanticscholar +1 more source
PAConv: Position Adaptive Convolution with Dynamic Kernel Assembling on Point Clouds [PDF]
Computer Vision and Pattern Recognition, 2021 We introduce Position Adaptive Convolution (PAConv), a generic convolution operation for 3D point cloud processing. The key of PAConv is to construct the convolution kernel by dynamically assembling basic weight matrices stored in Weight Bank, where the ...
Mutian Xu +3 more
semanticscholar +1 more source
Node-Feature Convolution for Graph Convolutional Networks [PDF]
Pattern Recognition, 2022 Graph convolutional network (GCN) is an effective neural network model for graph representation learning. However, standard GCN suffers from three main limitations: (1) most real-world graphs have no regular connectivity and node degrees can range from one to hundreds or thousands, (2) neighboring nodes are aggregated with fixed weights, and (3) node ...
Zhang, L., Song, H., Aletras, N., Lu, H.
openaire +1 more source
DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs [PDF]
IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016 In this work we address the task of semantic image segmentation with Deep Learning and make three main contributions that are experimentally shown to have substantial practical merit.
Liang-Chieh Chen +4 more
semanticscholar +1 more source
Grid Graph Reduction for Efficient Shortest Pathfinding
IEEE Access, 2023 Single-pair shortest pathfinding (SP) algorithms are used to identify the path with the minimum cost between two vertices in a given graph. However, their time complexity can rapidly increase as the graph size grows.
Chan-Young Kim, Sanghoon Sull
doaj +1 more source